Step 2.Choosing spatial objects for accuracy assessment.
It can be seen from Fig. 1 that the accuracy
of four different types of spatial objects needs
to be assessed: O E2- points, lines and polygons
only shown on the map of one particular scale;
(~) Eu- the same point features; @ Fj- the line
and polygon features defined by the same points;
and @ F2- the lines and polygons defined by non-identical
points representing natural features. Obviously,
there is also an accuracy issue with attributes
of spatial objects.
Step 3:Accuracy assessment for C2. The techniques
previously proposed by Burrough (1986),
Chrisman (1987), and Goodchild (1995) can be
used for assessing the accuracy of attributes of the
spatial objects shown on composite maps [~'1~
Step 4: Accuracy assessment for E2. The accuracy of geographic features of that type is identical
to the accuracy of the original map layer displaying
those features.
Step 5 :Accuracy assessment for s and F~. The
method presented in this paper is designed specifically
for handling these cases. Our approach attempts
to estimate the variance of digital representations
of points on the original data layers and
derived composite maps.
Step 6 9A ccuracy assessment for s In this case,
a more general approach is to fit a new line on the
final map by the least squares or maximum likelihood
methods with different weights for spatial objects
shown on source maps of different scales [l~
The variance of fittest points of the new line in the
final output can also be easily obtained.
At Steps 5 and 6, the approach proposed here
can also be used to eliminate the spurious polygons.